Morphogenesis and maturation of the embryonic and postnatal intestine

Abstract

The intestine is a vital organ responsible for nutrient absorption, bile and waste excretion, and a major site of host immunity. In order to keep up with daily demands, the intestine has evolved a mechanism to expand the absorptive surface area by undergoing a morphogenetic process to generate finger-like units called villi. These villi house specialized cell types critical for both absorbing nutrients from food, and for protecting the host from commensal and pathogenic microbes present in the adult gut. In this review, we will discuss mechanisms that coordinate intestinal development, growth, and maturation of the small intestine, starting from the formation of the early gut tube, through villus morphogenesis and into early postnatal life when the intestine must adapt to the acquisition of nutrients through food intake, and to interactions with microbes.

Introduction

The mature intestine is a highly complex organ with several essential functions. The small intestine interacts with food after it has been digested in the stomach and broken down into simpler units. Carbohydrates, proteins, lipids, and other nutrients are absorbed by intestinal enterocytes and are absorbed into a highly integrated vascular network.

In addition to absorbing nutrients, the intestine has important roles in host immunity. Within the intestine, luminal contents come into contact with an epithelial layer, which must serve as a barrier to the outside environment and protect the body against indigenous (commensal) microbes and pathogens. Critical to this barrier are epithelial tight junctions which selectively limit the passage of luminal contents in between epithelial cells [1]. In addition, the epithelium secretes mucus, which lines the intestinal tract and serves as a dense barrier that can trap microbes to inhibit infection [1], and can also provide a rich source of nutrients for commensal bacteria [2,3]. Specialized cells of the intestinal epithelium also play an important role in host immunity by secreting antibacterial and antifungal peptides [4–6]. Moreover, colonization by commensal bacteria at birth stimulates immune system development and is necessary for proper immune function [7].

To adequately fulfill the cellular demands of these complex functions, the intestinal epithelium is organized into villi, which are finger-like structures that protrude into the lumen. The intestine has a high rate of epithelial cell turnover, driven by proliferative epithelial stem cells housed at the base of the villi in domains called crypts (Figure 1). Stem cell driven proliferation fully regenerates the intestinal lining every 5–7 days [8–13]. As these stem cells divide, they differentiate and move along the villus structures like a conveyor belt. Once they reach the villus tips, cells undergo programmed cell death and slough off into the lumen.

Figure 1. The adult epithelial crypt-villus unit.

The adult intestinal epithelium is arranged in crypt-villus units. Intestinal stem cells (ISCs) and Paneth cells are housed in the crypt. A Transit Amplifying (TA) zone is a site for rapid proliferation and amplification of undifferentiated progenitor cells as they begin to make cell fate choices. Differentiated cell types continue to move up the villus in a conveyer-belt fashion where they carry out their day-to-day function, until they reach the villus tip where they undergo apoptosis and slough off into the lumen. Villus cell types include enterocytes, goblet cells and enteroendocrine cells, as well as enterochromafin cells, tuft cells and M-cells (not shown).

The highly archetyped crypt-villus structures of the adult intestine emerge over developmental time through the coordination of several complex processes that govern tissue patterning, cell fate, and morphogenesis. Early in embryonic development, the intestinal epithelium is a relatively flat, tube-shaped structure which undergoes a process called villus morphogenesis through which the relatively flat tube-shaped intestine gives rise to villi and crypt structures. Villus structures project into the lumen, expanding the apical surface area of the absorptive epithelium. Morphogenesis of these projections is an evolutionarily conserved process, and therefore a positive adaptation of fitness, found in many vertebrates including the chicken and mouse, but also in vertebrates such as zebrafish, seahorses, snakes, and amphibians [14,15]. This morphogenesis is responsible for a massive expansion of intestinal surface area; it is estimated that the absorptive surface area of the adult human intestine is 30 m², with villus structures amplifying the surface area by 6.5 fold [16]. Abnormal loss of absorptive surface area hinders nutritional uptake and can lead to malabsorption or intestinal failure [17].

In this review, we will discuss the molecular, biochemical, and biophysical events that guide normal intestine development and functional maturation of the postnatal gut, with a focus on mammalian development including human intestinal development where possible. To further focus the review, we will cover developmental events starting after gut tube formation and through early postnatal life.

Models of the developing intestine

Historically, many studies of vertebrate intestinal development have been carried out in the chick and mouse. Chick embryos are easy to acquire, develop rapidly, are low cost, and can be easily manipulated experimentally. However, tools for genetic manipulation in a tissue specific manner are more limited in the chick. Additionally, there are significant differences between avian and mammalian gut development that may limit cross-species comparisons [18]. The embryonic mouse model develops in a similar time frame to the avian embryo (19–21 days), and as an advantage, has an extensive set of tools for tissue specific genetic manipulation. Drawbacks include larger housing costs, long breeding times to obtain genetic crosses, relatively small litter sizes, and internal development which hinders experimental access to the developing tissue. Most importantly, it is not entirely clear which aspects from these models may be directly applicable to human intestine development, since our understanding of human intestine development is severely limited at this time.

However, access to human fetal tissue and in vitro tissue culture techniques using human pluripotent stem cells (hPSCs) have begun to shed additional light into human intestine development. Indeed, recent access to high-resolution 3D-reconstructions of early stage human embryos in addition to histological sections will likely improve our understanding of the early stages of human fetal gut development [19]. However, most studies of human fetal tissue are limited to descriptive analyses. hPSCs, which include both embryonic and induced pluripotent stem cells, represent a highly tractable solution to the limitations inherent to fetal tissue. hPSCs can be differentiated into complex 3-dimensional (3D) intestinal tissue using soluble growth factors and/or small molecules in a step-wise process known as directed differentiation [20–22]. Directed differentiation aims to recapitulate key developmental stages in vitro. In the case of intestinal tissue, hPSCs undergo a gastrulation-like process that gives rise to a mixed endoderm/mesoderm population, followed by posterior patterning events, intestinal specification and gut-tube morphogenesis which gives rise to small self-assembling 3D structures that can be expanded into ‘organoids’ [23–26]. Intestinal organoids have been reviewed extensively elsewhere [22,27–31].

Recent studies have shown that intestinal organoids derived from hPSCs are most similar to fetal intestine [23,27,29,32] [33]. Intestinal organoids transplanted into the mouse kidney capsule engraft, form villus and crypt structures, and undergo enhanced cellular, molecular and structural maturation, resulting in more adult-like tissue [27,32]. In addition to hPSC-derived organoids, in vitro culture of primary human fetal intestinal epithelium (fetal organoids) is also shedding light on the cellular dynamics of the human fetal intestine [34]. Collectively, hPSC-derived organoids and fetal organoids provide a powerful new platform for investigating human development, since both systems are experimentally tractable, allowing for long- term growth, and genetic and pharmacologic manipulation.

Intestinal specification, gut tube patterning, and formation

In the case of human gastrulation, like the chick, the endoderm, mesoderm and ectoderm lineages are specified and are present as a flat, layered disc-shaped structure (reviewed elsewhere: [23,35–38]) As development progresses, the body of the embryo rotates from a flat to a fetal position where the ectoderm is present on the outside of the embryo and the endoderm, wrapped by mesoderm, is present on the inside of the embryo [39]. Conceptually, the endoderm can be visualized as a flat sheet of paper that is folded into a tube that must be sealed in the middle as the two sides come together. In the mouse, gut tube closure is complete by E9.0 [23,39], but mutant mice lacking Gata4, Sox17, and Furin/SPC1 fail to rotate properly and have open gut tubes [40–44].

During embryo rotation and coinciding with complex morphological events that shape the tissue, the nascent gut tube is patterned into different domains along the anterior-posterior axis. Secreted morphogens help to establish region-specific gene regulatory networks, segmenting the gut tube into domains with distinct molecular characteristics that will ultimately give rise to different organs [45–47]. This process is reviewed in detail elsewhere [20,22,23,35,39,48–52]. For example, the foregut and hindgut domains of the endoderm are separated by expression of Sox2 and Cdx2, respectively [48,49,53,54]. The anterior region of the gut tube, which gives rise to the esophagus and stomach in addition to the lungs, liver, and pancreas, initially expresses Sox2, which sets up a sharp boundary at the pylorus [48,55]. Adjacent to this Sox2 boundary is the posterior region of the gut tube, which will give rise to the small and large intestine, marked by Cdx1, 2, and 4 expression [35,48,54,56–60].

Interestingly, while Cdx (Cdx1, 2, 4) proteins have been shown to play redundant roles in intestinal patterning during development [61–63], Cdx2 is considered to be a master regulator of intestinal identity; conditional deletion of Cdx2 in the epithelium resulted in complete loss of the intestinal gene expression program as well as loss of intestinal structure [54,60]. In these mutants, the gut tube formed normally; however, mutant tissue adopted an esophagus-like fate suggesting Cdx2 is absolutely required for intestinal commitment. Conditional deletion of Cdx2 later in development, around E13.5, resulted in transformation of the intestine into stomach-like tissue [60]. Collectively, these studies suggest that Cdx2 is critical for not only for specification but also for maintenance of intestinal identity during development. Interestingly, in the adult loss of Cdx2 does not lead to homeotic transformations, but instead impairs enterocyte differentiation, suggesting that Cdx2 affects intestinal identity only in the developing embryo [63,64].

Wnt signaling is critical for mid- and hindgut development, and plays a central role in inducing Cdx2 gene expression and specifying the intestinal endoderm both in vivo and in vitro [23,46,49,65,66]. Several studies have shown that Wnt/β-catenin signaling is present at higher levels on the posterior side of the developing embryo and several Wnt ligands are highly expressed in this region [46,67–72]. In addition, Wnt/β-catenin signaling is active for longer durations in the posterior of the embryo as it elongates. This sets up both temporal and spatial gradients within the embryo, where the developing midgut endoderm is exposed to Wnt signaling at lower levels and for shorter periods of time whereas hindgut endoderm is exposed to Wnt signaling at higher levels for a longer period of time. There is emerging evidence that these early signaling gradients in the embryo endoderm may help to establish intestinal regional identity, setting up the different domains of the intestine: the duodenum, jejunum, ileum and colon. This notion is supported by studies using mouse embryonic explants. Stimulating endoderm with high levels of Wnt signaling led to induction of posterior small intestine and colonic gene expression in the endoderm [49]. Interestingly, stimulating mouse embryonic stem cell (mESC) derived endoderm with high levels of Wnt signaling induced a Cdx2-positive small intestinal fate, but was unable to induce a colonic fate, suggesting that other factors may cooperate with Wnt signaling to drive colonic specification [49]. In line with these studies in mouse embryos and mESCs, recent studies using hPSC-derived intestinal organoids demonstrated that the duration of Wnt/β-catenin stimulation in hPSC-derived endoderm cultures was associated with intestinal patterning, with shorter durations specifying proximal small intestine-like organoids (duodenum) and longer durations specifying distal small intestine-like organoids (ileum). Interestingly, and similar to mESC-derived endoderm, colonic gene expression was not induced in these studies, suggesting that colonic specification may require additional signals [73].

Early intestine development

By E9.5 in mouse development the Cdx2+ gut tube becomes a simple pseudostratified epithelium [23,74]. From E9.5 to E13.5 in the mouse, the epithelium and mesenchyme rapidly proliferate, resulting in elongation of the gut tube and increased intestinal length, circumference, and luminal area [75,76]. This period of embryonic development (E9.5–E13.5 in mice; approximately 3–7 weeks in humans) is the most poorly understood stage of intestinal development, likely due to the difficulty in analyzing tissue and embryos at these early time points.

It is known, however, that the increase in intestinal length/girth during early development is mediated in part by noncanonical Wnt signaling through the planar cell polarity (PCP) pathway [77]. Misregulation of Wnt5a signaling leads to defects in gut lengthening. Studies on Wnt5a null mice demonstrated significantly shorter gut tubes with bifurcation of the duodenum and perturbed midgut elongation as well as a truncation at the cecum [76]. Defects were apparent by E10.5, at the onset of midgut elongation, and corresponded to reduced epithelial proliferation and failure of epithelial cells to re-intercalate after mitosis [76], which is known to take place via interkinetic nuclear migration in the pseudostratified epithelium [74]. Similarly, mice lacking Secreted Frizzled Related Proteins (Sfrp), inhibitors of Wnt5a, display shortened guts with ectopic clumps of epithelia that protrude into the lumen at E13.5. Epithelial clumps displayed aberrant localization of aPKC, β1-integrin, and E-cadherin, indicating defects in apicobasal polarity [78] Notably, improper cell intercalation in frogs also results in gut lengthening defects [79]. Additionally, the Hh signaling pathway is required for normal intestine lengthening during early intestinal development. Conditional epithelial-specific deletion of Ihh by E10.5 resulted in loss of mesenchymal proliferation and dramatically shortened intestines. E12.5 Ihh-deficient intestines were 10% the length of their control counterparts [80].

Also taking place during this early developmental time is formation of the smooth muscle layers, which surround the gut tube to provide structure and later aid in peristalsis (reviewed in [81]). Smooth muscle differentiation starts around E11 in the mouse and proceeds in a proximal to distal wave along the length of the intestine. At E12, a layer of mesenchymal cells become circularly arranged and forms a distinct layer of Alpha Smooth Muscle Actin (αSMA) expressing circular muscle by E13 [82–84]. Over the next 48 hours, three distinct layers of smooth muscle are patterned—the circular smooth muscle and longitudinal smooth muscle of the muscularis propria, and the longitudinal smooth muscle of the muscularis mucosae [81]. Smooth muscle differentiation is dependent on Hh signaling from the epithelium. Hh ligands, Shh and Ihh, are expressed in the epithelium and signal to Ptc1 and Gli1 expressing mesenchymal cells at early developmental stages [85,86]. Classical experiments conducted in chick and mouse demonstrate that overexpression of Shh expands gut mesoderm and induces smooth muscle differentiation [87–89]. Mice deficient in Shh or Ihh display a 20–30% reduction in thickness of the circular smooth muscle layer at E18.5 [86]. Epithelial-specific conditional deletion of Ihh results in the loss of smooth muscle actin (SMA) expressing cells [80,90]. A gain of function study expressing constitutively active Smoothened in early gut mesenchyme at E9.5 demonstrated that ectopic activation of Hh signaling resulted in an expansion of SMA expressing cells, indicating that Hh drives expansion of smooth muscle cell progenitors [80]. Additionally, BARX1, a homeodomain transcription factor, specifies stomach cell fate and promotes proliferation of stomach smooth muscle cells. Ectopic expression of BARX1 in the intestine produces smooth muscle with gastric muscle morphology, malrotation and shortening of the gut [91].

Once the smooth muscle has formed, muscular contractions controlling peristalsis are coordinated by the enteric nervous system [92,93]. The gut tube becomes innervated upon migration of vagal neural crest cells starting at E9. These neural crest cells proliferate and migrate caudally throughout the myenteric region and later populate the submucosa. Around E14, neural crest progenitors give rise to sensory and motor neurons, which project nerve fibers into the gut, allowing colonization of Schwann cell precursors. Neurons and glial cell differentiation occurs and continues postnatally [81,94]. The detailed molecular mechanisms surrounding the enteric system in the gut exceeds the scope of this review and are reviewed elsewhere [81,94–96], but of note the RET/GDNF signaling pathway is perhaps the best characterized. Absence of RET/GDNF signaling abrogates the migration and differentiation of enteric neural crest cells and leads to enteric nervous system disorders like Hirschsprung’s disease [97,98]. Also of note, tissue engineered systems using hPSC-derived intestinal organoids and/or neural crest progenitors are now being implemented to better study human mutations that lead to innervation defects, causing improper gut function and dysmotility at birth [99,100].

Simultaneous with ENS development, the early intestine becomes vascularized. Pecam+ endothelial cells are present in the gut by E9.5 [93]. By E11 in the mouse, the serosal mesothelium begins to form on the surface of the gut and covers the peritoneal cavity [93], and at E12.5, mesothelial cells undergo EMT and enter the submesothelial space of the gut and over the next few days, they differentiate into vascular smooth muscle of the newly forming vascular network of intestinal arteries and veins [101].

Rapid proliferation of the epithelium also takes place in the early intestine, and is required for intestinal lengthening and expansion of the surface area. Prior to villus morphogenesis at E14.5, the pseudostratified intestinal epithelium is uniformly proliferative, but upon the emergence of villus architecture, epithelial proliferation becomes restricted to the intervillus domains [102]. It is well documented that proliferation in the intervillus domains that emerge following villus morphogenesis (starting around E15.5 in mice) and in the crypts of the adult intestine, is highly dependent on Wnt/β-catenin signaling [102–107]. However, recent studies have suggested that epithelial proliferation in the pseudostratified epithelium prior to villus formation is regulated by mechanisms independent of Wnt/β-catenin signaling. Target gene expression and Axin2-LacZ reporter mice suggested that Wnt/β-catenin signaling activity was low at E13.5 and dramatically increased over developmental time in the intestinal epithelium [108]. Supporting the notion that Wnt signaling is low in the pseudostratified epithelium, conditional epithelial deletion of β-catenin, or the Wnt co-receptors, Lrp5 and Lrp6, had little effect on epithelial proliferation in the E13.5-E14.5 mouse intestine, but led to a dramatic loss of proliferation after villus formation [108].

Similarly, mice null for Tcf4 (Tcf7l2), which is a transcriptional binding partner of β-catenin and is required for β-catenin dependent Wnt signaling, did not display proliferation defects in the pseudostratified epithelium, but completely lost epithelial proliferation after villus formation [102]. Collectively, these studies suggest that Wnt/β-catenin signaling is dispensable for proliferation during pseudostratified intestinal development. In this context, it is interesting to note that several separate studies have shown that the Wnt/β-catenin target gene, and well described adult intestinal stem cell marker, Lgr5 [13], is expressed during this time of low Wnt activity [109–111]. Recent lineage tracing experiments in Lgr5-creER mice have shown that lineage tracing can occur as early as E12.5 [109]. Mechanistically, it appears that the transcription factor Id2, restricts Wnt activity during this window of development and Id2-deficient intestinal epithelial tissue had more Lgr5+ cells starting at E9.5 compared to controls [111]. Evidence also suggested that Id2 deletion increased Wnt/β-catenin signaling activity in these animals. Collectively, these studies point to an interesting and unexplained paradox. While Lgr5 is considered a sensitive Wnt/β-catenin target gene in the adult intestine, it appears that it is already expressed during a time when Wnt/β-catenin signaling is very low in the fetal gut (E12.5–E13.5) [108–111]. On the other hand, removing Id2, which presumably lead to an increase in Wnt/β-catenin signaling, increased the number of cells expressing Lgr5 [111]. Thus, it is unresolved if, how and why Lgr5 is present when Wnt-signaling is very low but still seems to respond as a target gene when Wnt-signaling is activated in Id2-null epithelium. In addition, given that Id2 is a Bmp target gene in other systems [112,113], this work would suggest that Bmp is highly active in the epithelium during the pseudostratified stage of development; however, this idea has not yet been experimentally tested.

In addition to Id2 mediated repression of Wnt//β-catenin signaling, another mechanistic explanation for the increased Wnt/β-catenin signaling over time is that mesenchymal Wnt ligand production increases around E14.5. Supporting this notion, inhibition of mesenchymal Wnt ligand secretion by conditional deletion of Wntless in the mesenchyme results in loss of epithelial proliferation at E15.5 [108]. The signaling mechanisms that drive epithelial proliferation during the pseudostratified stages remain unknown. One candidate regulator of proliferation prior to villus development is Gata4. ChIP-Seq of adult mouse intestinal epithelia shows that Gata4 binds to many cell-cycle genes [114]. Additionally, conditional epithelial deletion of Gata4 disrupts epithelial cell proliferation by E10.5, resulting in delayed villus morphogenesis [114]. It is interesting to speculate that Gata4 may act through retinoic acid signaling to regulate proliferation as it has been shown to be a downstream target of retinoic acid in the intestine and other endoderm derived tissues [115,116].

Villus formation

Villus morphogenesis is a process where the flat pseudostratified intestine begins to remodel and give rise to villus structures, which consist of finger-like epithelial protrusions into the intestinal lumen with an underlying mesenchymal core [18,23]. Villus formation massively expands the intestinal epithelial surface area, allowing for sufficient nutrient absorption to sustain life. As such, villus morphogenesis is a complex process that is driven by a combination of inductive cues and physical forces, which coincide to coordinate this morphological process [18]. Individual villi are connected to neighboring villi by proliferative intervillus domains (also called intervillus zones) (Figure 2). Within the past 5 years, a plethora of work in the mouse and chick has shed significant light on the regulation of villus morphogenesis, and has highlighted significant species-specific differences in this process [14,74,84,110,117]. In addition, recent studies of human fetal development revealed that villi begin to form between 51–54 days of gestation correlating to the beginning of villus morphogenesis at E14.5 in the mouse embryo [19,117,118]. In mice, villi emerge in a proximal to distal wave, arising first in the duodenum and spreading to the ileum over a span of 36 hours [117], and this trend appears to be consistent in the human fetal intestine [19]. Interestingly, for many years it was thought that the human and mouse intestine initially formed micro-lumens in the flat epithelium, which then went on to fuse, giving rise to villi [23]; however this was recently shown not to be the case in mice, and 3-dimensional reconstructions demonstrated that the lumen was continuous during villus formation [74]. Interestingly, new data from the ‘three-dimensional digital atlas and quantitative database of human development’ [19] has shown that micro lumens may be present in the developing human intestine representing an interesting species-specific difference, although it should be noted that more detailed follow up studies will be needed to definitively show any potential differences (Figure 3).

Figure 2. Developmental epithelial transitions and mesenchymal cluster formation in the mouse intestine.

A. The early murine intestinal epithelium (yellow), between E9.5-E13.5, is present as a flat pseudostratified epithelium within the gut tube. B. Beginning around E14.5, mesenchymal clusters (red) aggregate adjacent to the epithelium where a nascent villus will form. Cluster formation causes a deformation in the epithelium above the cluster. C. Villi form above the cluster, establishing the highly proliferative intervillus domain between villi. Several rounds of villus morphogenesis will occur, and new clusters will form (blue) adjacent to the intervillus domain following completion of the prior round of cluster-villus formation (red clusters).

Figure 3. Human fetal intestine development.

Sections through different Carnegie Stages (CS) of the developing human embryo were obtained (http://www.3dembryoatlas.com and de Bakker et al., 2016) and traces of the proximal small intestine (duodenum) were generated. The intestinal epithelium (yellow) appeared to have multiple lumens prior to villus morphogenesis (CS18), and nascent villi formation was apparent by CS20. Image resolution was not sufficient to determine if the human intestine formed villus clusters in the mesenchyme (red). Villus structures became more pronounced, and greater in number as development progressed (CS21–CS23).

While little-to-nothing is known mechanistically about villus formation in humans, it is well appreciated in the mouse that signaling molecules secreted from the rapidly proliferating pseudostratified epithelium act as critical regulators of villus morphogenesis. Hedgehog (Hh) and Platelet Derived Growth Factor (Pdgf) signaling are well established signaling pathways that regulate this process. Epithelial Hh and Pdgf ligands signal to the underlying mesenchymal cells, which express the pathway receptors Ptch1 and Pdgfra, respectively [85,117–119]. As Hh and Pdgf signaling is activated in the mesenchyme adjacent to the epithelium, it stimulates these cells to exit the cell cycle and aggregate into small dense clusters [85,117–119]. Cluster formation coincides with the initiation of a nascent villus in the epithelium overlying the cluster. While the cluster itself expresses several signaling molecules, including Bone Morphogenetic Protein (Bmp) ligands, it is not clear how the epithelium-cluster unit initiates the formation of a nascent villus [84,118]. Nonetheless, formation of the cluster is an absolute prerequisite for villus formation, since mutations in the Hh or Pdgf pathways perturb mesenchymal cluster formation and disrupt subsequent villus formation, with Hh signaling being the most critical to this process as blocking Hh signaling can completely block mesenchymal clustering and villus formation [80,84,117,119]. Alternately, increased Hh signaling in explant cultures by the addition of a pathway agonist (Smoothened agonist; (SAG)) increased the size of cluster and villus structures [117].

Although mesenchymal clusters form and express Bmp ligands, it does not seem that mesenchymal Bmp’s immediately signal back to the epithelium, since genetic deletion of Bmp receptors in the overlaying epithelium does not lead to perturbations in villus formation [84]. In addition, mesenchymal clusters also express inhibitors of Bmp signaling, including Noggin (Nog) and Twisted Gastrulation 1 (Twsg1), and functional experiments have shown that perturbing Bmp signaling affects the size and patterning of the mesenchymal clusters [84]. These functional experiments have led to the hypothesis that Bmp signaling establishes the regular spacing and patterning of mesenchymal clusters in an activator-inhibitor reaction-diffusion style mechanism [84].

The reaction-diffusion model that may explain the distribution and patterning of mesenchymal clusters in the intestine as recently suggested by Walton et al., [84] was first proposed by mathematician Alan Turing, who described a model where an activator and inhibitor emanating from the same source interact to establish a self-organized and predictable pattern [120]. It is interesting to note that in his manuscript, Turing noted, “…the description of the state consists of two parts, the mechanical and the chemical”. He then goes on to state, “One cannot at present hope to make any progress with the understanding of such systems except in very simplified cases. The interdependence of the chemical and mechanical data adds enormously to the difficulty, and attention will therefore be confined, so far as is possible, to cases where these can be separated” [120]. Thus Turing acknowledged, but did not address, the mechanical forces that would normally be present in a biological system. In this light, it is interesting to note the work of Oster and colleagues many years later, who used mathematical modeling of mesenchymal cell behavior to propose that mechanical traction forces exerted by the mesenchyme on the surrounding extracellular matrix could deform the matrix and affect both the direction of mesenchymal movements and the formation of pattern [121–123]. In this work, it was proposed that mesenchymal traction forces would eventually lead to a uniform distribution of cells breaking up into local cell condensations, characterized by “islands of high cell density alternating with regions of low cell density” [121]. Oster’s model also predicted that as tissues grow and mature, developmental waves of mesenchymal condensations could form behind maturing tissue in regular patterns [121]. For example, it was predicted that an anterior-to-posterior gradient of mesenchymal condensations could form in a developing tissue as a population of cells became developmentally competent to form cell aggregates in an age/time dependent manner [121]. It is interesting to speculate that such traction forces could cooperate with morphogen signals, and that both may play a role in the formation or propagation of the anterior-posterior wave of mesenchymal clusters that condense during villus morphogenesis, since it is well described that the intestine matures in an anterior-posterior fashion. At this point in time there is no evidence in the developing mouse intestine to show how signaling and biomechanical forces may cooperate during the process of mesenchymal clustering in mice.

On the other hand, it has been predicted that tension/force placed on the epithelium by the underlying mesenchymal clusters may instruct the overlying epithelium to start forming a villus [124]. Here, it has been proposed that nascent clusters deform the epithelium as they form on the basal side of the epithelium, placing compression forces on the epithelium and placing strain on the apical epithelial surface. When coupled with reduced tension in the apical F-actin cortical network due to a mitotic event in the highly proliferative epithelium, sufficient strain from below the epithelium coupled with a local reduction in apical surface tension would lead the epithelium to deform and buckle in between the mesenchymal clusters, effectively forming nascent villi above the clusters [124].

Recent work has also linked tensile forces produced by radial smooth muscle on the development of villus structures in the developing chick gut [14,110]. This work demonstrated that as the smooth muscle layers differentiate, they place a global compressive force on the intestine, effectively stiffening the outer wall of the intestine. As the highly proliferative epithelium continues to expand in a uniform manner circumferentially, the compressive forces created by the smooth muscle cause the epithelium to buckle. As each subsequent muscle layer differentiates in the developing gut – the circumferential muscle layer, followed by the exterior longitudinal muscle and then the interior longitudinal muscle – new mechanical strains are placed on the intestine, leading to an initial pattern of epithelial ridges, followed by a zig-zag pattern and finally, villus structures [14]. It is interesting to note that there are significant differences between chick and mouse intestine development. For example, proliferation and expression of the Wnt/β-catenin target gene, Sox9, is rapidly localized to the intervillus domain following villus formation in the mouse [108], whereas the chick epithelium continues to be broadly proliferative and expresses Sox9 throughout the epithelium following villus formation [14,110]. In the chick, the epithelial folding induced by mechanical constraint is suggested to help to concentrate morphogenetic signals in the underlying tissue. For example, epithelial Hh ligands are concentrated in the underlying mesenchyme after epithelial buckling to induce changes in the mesenchyme [110]. Collectively, these data suggest that the mechanisms driving villus development and regulating epithelial and mesenchymal differentiation in the two species are dramatically different, but also show the importance of both signaling molecules and biomechanical forces in tissue morphogenesis in both species [14,84,110,124].

Development and maturation of the intestine following villus formation

Following villus formation, epithelial proliferation becomes rapidly restricted to the intervillus domains at the base of the villi for the remainder of development [108,125,126]. Two major signaling pathways that are important for intestinal proliferation in the late neonatal and adult intestine are the Wnt/β-catenin and Notch signaling pathways [106,127–134]. For detailed reviews on Wnt and Notch signaling in the intestinal stem cell, see: [135,136]. Although proliferation in the pre-villus intestinal epithelium in mice can occur in the absence of Wnt/β-catenin signaling, once villi form at E15.5, proliferation is dependent on Wnt/β-catenin signaling. Blocking signaling activity using a variety of genetic methods results in a complete loss of epithelial proliferation [108,125,137–139]. In addition, several Wnt/β-catenin target genes become restricted to the proliferative intervillus domain following villus morphogenesis, including Axin2, CD44, CyclinD1, Sox9 and Lgr5, reinforcing the location of Wnt/β-catenin signaling activity following villus formation [55,108,110,125,139,140]. Interestingly, while Notch signaling is a critical regulator of intestinal stem cells postnatally and in the adult [127,130,141,142], the role that Notch signaling plays at earlier times of development is not clear. Notch genetic gain-of-function studies have been conducted in the developing intestinal epithelium, and have shown that developmental misregulation of Notch Intracellular Domain (NICD) expression can lead to increased epithelial proliferation [143], or can lead to a block in proliferation [144]. These opposing results are likely explained by the different Cre drivers used and the developmental timing of NICD expression. Recent studies in the developing mouse intestine, the human fetal intestine, and in hPSC-derived intestinal organoids have shown that the Notch target gene, OLFM4, is expressed at extremely low levels relative to the adult intestine, suggesting that Notch signaling may be less important for regulation of the progenitor cells in the embryonic/fetal intestine, although this idea has not been experimentally tested [27,34].

Shortly after villus emergence, around E16.5 in mice, the epithelium on the villi begins to undergo cytodifferentiation in to the functional cell types of the small intestine, including secretory cells - mucus producing goblet cells and hormone-producing enteroendocrine cells –and absorptive enterocytes, which comprise more than 80% of all intestinal epithelial cells and are responsible for absorbing nutrients from the lumen [23,128,145,146]. A detailed review of the molecular mechanisms controlling differentiation in the intestinal epithelium is outside the scope of the current article, but readers are encouraged to see the following reviews: [145–148]. In brief, Notch signaling is known to play an important role regulating the choice to differentiate into a secretory cell (Notch OFF) or into an absorptive enterocyte (Notch ON) [129,146,149–151]. While most of our understanding about cellular differentiation in the intestine has been established through studies in the postnatal intestine, in the developing gut Gata4 and Gata6 function redundantly to suppress proliferation and regulate cytodifferentiation of goblet cells by modulation of Notch signaling [152]. In addition, the transcription factor, Klf5, is required for initiation of differentiation in the developing gut, as genetic deletion of Klf5 from the intestinal epithelium impaired differentiation [126].

In mice, the crypt of the intestine emerges around postnatal day 14. The mechanisms by which the embryonic intervillus domains give rise to the postnatal/adult crypt are completely unknown. However, crypt emergence coincides with differentiation of Paneth cells in the intestine [153,154]. In humans, Paneth cell differentiation occurs around week 20 of fetal gestation [155,156]. Paneth cells initially emerge at the 5–7^th cell position in the crypt and then migrate downwards to the base of the crypt adjacent to LGR5+ stem cells [154,157]. Paneth cells secrete defensin proteins [158–160], which are known to have antimicrobial properties to protect against pathogen infection and also play a role as a niche cell, supporting intestinal stem cell maintenance [4–6,161]. Paneth cell differentiation is initially controlled through a Notch dependent mechanism during secretory progenitor specification and further Paneth cell maturation is regulated by Wnt signaling [105,129]. Deletion of Lgr5 in the embryonic intestine led to increased levels of Wnt signaling and precocious Paneth cell differentiation [137]. Further differentiation of Paneth cells requires the expression of the Wnt target Sox9 [140,162] and mice with conditional deletion of Sox9 lack Paneth cells in the crypts [163,164].

One of the hallmarks of intestinal epithelial maturation is the acquisition of fully functional epithelial cell types [27,34,165] and reviewed in [166]. In the human fetal intestine and in hPSC-derived intestinal organoids, this includes the differentiation of Paneth cells and many enzymes that function in nutrient absorption that are present on the enterocyte brush boarder [27]. In the mouse, epithelial maturation is regulated by the transcriptional repressor Blimp1 (also known as Prmd1) [167,168]. Blimp1 is expressed broadly throughout the intestinal epithelium of the embryonic intestine, with expression becoming restricted shortly after birth, when it is excluded from the proliferative domain (intervillus domain and emerging crypt) over the first 2 weeks postnatally. By the third week of life and through adulthood, Blimp1 is expressed only in the tip of the villus. Blimp1 mutant mice show early differentiation of Paneth cells, an increased differentiation of goblet cells, and a major metabolic shift towards the adult phenotype by postnatal day 7 at the expense of suckling-period-specific enzymes [167,168]. Mechanistically, ChIP-seq experiments showed that Blimp1 is able to bind to DNA associated with metabolic genes in the epithelium adding evidence to the notion that Blimp1 repressed gene expression adult expressed genes during the embryonic period [169]. Further, it was shown that Blimp1 was bound to many of the same regions as the transcriptional activator, Irf1. Irf1 was bound and can activate transcription of MHC class I pathway genes, and it was postulated that an additional role for Blimp1 was to repress Irf1 bound genes while the neonatal gut acquires tolerance during microbial colonization over the first few weeks of life [169].

Recent work has also suggested that major shifts in the metabolism of the developing intestinal epithelium play a significant role in maturation of the intestinal epithelium [170]. This study found that there was an increase in the expression of genes involved in oxidative phosphorylation coincident with villus development, and that expression of these genes continued to increase throughout embryonic development. It is interesting to note that metabolism in the adult intestine shifts from glycolysis-to-oxidative phosphorylation along the crypt-villus axis [171], indicating that this developmental switch may be critical in preparing the epithelium for postnatal life. The embryonic shift in oxidative phosphorylation genes was controlled by the transcription factor Yy1, as genetic deletion of Yy1 in the epithelium led to reduced gene expression and stunted villus growth. Further supporting the connection between oxidative phosphorylation and intestinal growth, pharmacological inhibition of the electron transport chain caused a similar stunting of villus growth. Interestingly, oxidative phosphorylation was reduced in human neonates with necrotizing enterocolitis (NEC), indicating that this metabolic shift may be critical for the intestine to mature and adapt to neonatal life [170].

Postnatal acquisition of the gut microbiome and education of the immune system

While the cellular and structural features of the gastrointestinal tract are largely in place at birth, functional maturation and adaptation to the extra-uterine environment requires colonization by mutualistic microorganisms and the coalescence of a functional microbial community that is fully integrated with the nutritional and immunological requirements of the human host [172,173]. Germ-free mice exhibit profound developmental defects [7] including severely impaired adaptive immune development [173,174], altered systemic metabolism [175], decreased epithelial turnover and impaired villus formation [176], altered glycosylation of the luminal surface of the intestine [177,178], and increased epithelial barrier permeability [179]. Disruptions in gut bacterial colonization and subsequent immune development caused by antibiotic exposure [180], Caesarean birth [181], or prematurity [182–184] are associated with autoimmune and inflammatory diseases [185,186], developmental delay [187], and infection risk [188] during the neonatal period and beyond [185,189]. Colonization of the gastrointestinal tract occurs during birth, feeding, and through exposure to the extra-uterine environment. This exposure aids in the development of a healthy immune system.

An innate system of host defense proteins is critical for containing commensal and pathogenic microbes during initial colonization of the intestine. The first line of defense includes mechanical, chemical, and microbiological barriers of the epithelium, such as tight junctions (mechanical), mucous (mechanical), secreted antimicrobial peptides (chemical and microbiological), and indigenous microbes (chemical and microbiological) [190]. Inadequate maturation of these barriers, as is the case in preterm infants, creates a vulnerability to microbiological invasion and can lead to severe and life threatening illnesses such as necrotizing enterocolitis [29,184].

Ensuring colonization of the gastrointestinal tract with appropriate microbial communities and functional adaptation to a vast diversity of dietary and microbial challenges requires a lengthy process of heuristic education [4,5]. An adaptive immune system within the mature intestinal mucosa contains an array of immune cells loosely organized as gut-associated lymphoid tissue (GALT) and consisting of lymphoid cells, (T and B cells) and myeloid lineage cells (macrophages, neutrophils, eosinophils and mast cells). GALT includes several clusters of lymphoid tissue including tonsils, Peyer’s Patches in the small intestine, and other smaller aggregates in the appendix, stomach, esophagus, and large intestine [191]. Peyer’s Patches first begin to appear in mice around E15.5 and in humans around 19 weeks of gestation [192]. This lymphoid tissue allows the immune system to develop a “memory” of previous antigen exposure that mediates tolerance to indigenous gut microbes and responds rapidly to eliminate potential pathogens.

Postnatally, GALT tissue also serves to “educate” the immune system. Highly specialized epithelial Microfold cells (M cells) are interspersed within the epithelium and mediate antigen presentation to lymphoid cells to induce antigen specific responses [193]. M cells are restricted to the mucosa directly overlying Peyer’s patches. M cells express numerous cytoplasmic vesicles, facilitating the transport of luminal antigens across the epithelial barrier and presentation at the basal epithelium to the underlying lymphoid cells [194,195]. Additional dietary, microbial, and self-antigens traverse the epithelial barrier through paracellular and transcellular pathways [1].

Dormant T-cells reside in tissues or circulate throughout the lymphatic system in the absence of antigen corresponding to the unique T cell receptor. Upon interaction with cognate antigen presented by dendtritic cells, M cells, or passing directly across the epithelial barrier, T-cells respond by proliferation and generation of distinct cytokine signals that activate immunoglobulin-producing B-cells with compatible antigen recognition properties [196]. B-cells within patches do not mature to an IgG producing state until several months of age. The fetus will receive the vast majority of its IgG from the mother transplacentally [190]. However; initial exposure to antigen during infancy ultimately results in the expansion of T-cells and B-cells with gut antigen specific lymphoid cells, enhancing recognition of the antigen and production of specific antibodies subsequently [196].

Mammalian evolution has confronted the challenge of postnatal nutrition and development over countless millennia, resulting in a transitional food, milk, which is produced in the mammary glands of all mammals and consumed by offspring throughout postnatal development. Milk contains a complex matrix of proteins, lipids, and carbohydrates shaped by natural selection to enhance the fitness of mammalian offspring through the provision of rich nutrition, developmental signals, and defense against pathogens. The comparison of breastfed and formula fed infants indicates that immune maturation and function are enhanced in the gastrointestinal tract by specific human milk components, such as IgA. The immature intestinal mucosa is prone to damaging inflammation [197], however breastfed infants are at a substantially reduced risk of developing inflammatory and autoimmune diseases such as necrotizing enterocolitis [198,199] and allergic disease [200–202] throughout childhood and are at significantly lower risk of developing inflammatory bowel disease in adulthood [189]. Early observations that milk contains many known inflammatory mediators and that maternal leukocytes present in milk have poor function led to the hypothesis that milk factors are anti-inflammatory per se [203]. The mucosal immune system is an integral component of intestinal physiology, regulating the dynamic passage of ingested material through the epithelial barrier in the presence of billions of commensal organisms and potential pathogens; dysregulation of these processes is now thought to be a major factor in the etiology of chronic inflammatory diseases of the GI tract, including necrotizing enterocolitis [204]. Breastfed infants exhibit rapid maturation of the intestinal barrier function relative to formula fed infants [205].

IgA is the major immunoglobulin of the intestinal mucosa and while undetectable at birth, is abundant in breastmilk and will populate the mucosal surface once feeds are initiated [206]. In humans, this passive immunity lasts several months to protect the infant after birth. Nursing mice deficient in IgA are protected from infection by enteric pathogens such as Salmonella typhimurium and Citrobacter rodentium, implying that immunoglobulins are only one part of the multi-faceted protection conferred by milk antimicrobials [207]. Other milk proteins, including lysozyme, lactoferrin, and α-lactalbumin exhibit non-specific anti-microbial properties [208] and may act to limit bacterial overgrowth or impose selective pressures that shape the composition of the gut microbiota in infancy [209]. Transforming growth factor β (TGF-β) is found at concentrations as high as 1.5 μg/mL in human milk [208,210]. TGF- β has diverse roles regulating immune activity as well as cellular differentiation and proliferation [211], and TGF-β in milk promotes attenuation of the immune response and the induction of antigenic tolerance [202,212]. In mouse models of intestinal inflammation, dietary TGF-β supplementation was associated with improvements in inflammatory pathology and weight retention [213]. Following several successful clinical trials [214,215] TGF- β supplemented formula is now widely used to treat pediatric Crohn’s disease.

Among the bioactive components of human milk, indigestible glycans, a group that includes oligosaccharides, glycosaminoglycans, glycolipids, and glycoproteins, exert a wide variety of protective functions in the developing gastrointestinal tract [216]. This includes prebiotic functions, immunomodulation, anti-adhesive pathogen inhibition, and enhancement of gastrointestinal function [208]. Indigestible oligosaccharides in milk play a key role as selective substrates for microbial growth, facilitating the rapid expansion of beneficial microorganisms within the infant gut to the exclusion of pathogens and other organisms [217,218]. Expression of specific oligosaccharide structures in milk varies widely between individuals and may account for some of the variation seen in infant gut microbiome community structure and disease risk [219].

Conclusion and future directions

The gastrointestinal tract is a highly complex and multifunctional organ system. Decades of work in multiple disciplines have resulted in a framework for our understanding of the development of the structures and functions that comprise the mature intestine.

Formation of an integrated, comprehensive understanding of the growth, development and maturation of the intestine and its dynamic function throughout life will require the continued use of multidisciplinary approaches and the implementation of new model systems that allow for new ways to further our understanding. Classical animal models such as the frog and chick have proven to be powerful experimental tools for understanding the cellular and genetic basis of gut development, and have allowed us to explore the molecular pathways important during gastrulation, formation of the endoderm and gut morphogenesis. Mice have also been an excellent model to study mammalian intestinal development. These model systems have provided insights into similarities and differences between species; most notably are the distinct mechanisms behind the extension of luminal surface area and formation of villus structures. However, there is still a large gap in knowledge regarding the mechanisms regulating development of the human intestine.

Recent developments, including access to information from human embryos, the ability to culture primary fetal intestinal tissue and the use of hPSC-derived tissues such as intestinal organoids, provide new and exciting avenues to understand human intestinal development. These tissues will facilitate additional research into the complex processes involved in tissue maturation, including how nutrients and microbes may influence human gut development. Our understanding of the development of the gut vasculature, enteric nervous system and immune system, and how these tissues influence development and maturation remains incomplete. Mechanistically, we are only beginning to understand how local mechanical forces interface with morphogens to control complex biological processes, and how broad forces such as those generated by peristalsis and luminal flow may also influence gut development. Lastly, metabolism may be a critical regulator of intestinal development and maturation but still remains unexplored. Studies investigating these questions are ongoing and promise to further our understanding of this fascinating organ system.